Pentagon AI chief says the technology could help spot future pandemics earlier
More money should be spent on information-age weapons, Air Force Lt. Gen. John N.T. ‘Jack’ Shanahan says as he retires from service
The Pentagon’s top official overseeing the military’s artificial intelligence efforts sees a significant role for the technology to track and spot pandemics in the future like COVID-19, a disease that has caused large-scale destruction of lives and economic distress similar to wars.
“I do believe there’s great potential to bring in artificial intelligence to provide early warning of future problems” such as disease outbreaks, Air Force Lt. Gen. John N.T. “Jack” Shanahan, director of the Pentagon’s Joint Artificial Intelligence Center, said in an interview.
Defense officials and technologists ought to be thinking about what such a “national warning network would look like” to understand how to assemble the right sources of data and build predictive models, he said.
A few private companies did in fact use artificial intelligence tools to provide early warning of the pandemic, Shanahan said, adding that predictive models should be refined to yield more accurate results.
BlueDot, a Toronto-based company that uses artificial intelligence to provide early warning on infectious diseases, did alert its clients as early as December 2019 about an unusual outbreak of pneumonia in Wuhan, China, about 10 days before the World Health Organization did. A Boston hospital and another company in San Francisco also are said to have spotted the early signs using AI tools.
Even if all the right sources are lined up and data is ingested, not many organizations and government agencies are “doing AI and machine learning after aggregating and infusion of data to provide some sort of predictive analytics,” said Shanahan, who began his career as an F-4 Phantom pilot and plans to retire next month after 36 years in the military.
Nand Mulchandani, the chief technology officer of the artificial intelligence center, will serve as its interim director until the White House names a military officer who then has to be confirmed by the Senate.
One of the key goals of the artificial intelligence center known by its acronym — JAIC, pronounced like Jake — is to change the U.S. military’s culture from the industrial-age emphasis on hardware and weapons to the digital-age emphasis on data. Shanahan helped set up the center in 2018 and became its first director. The United States and China are racing to gain an advantage in artificial intelligence-enabled tools that will allow their militaries to respond to threats at computer speeds rather than at human decision-making times.
As the pandemic spread, shutting down government agencies and companies across the world, the artificial intelligence center kicked off a new predictive modeling effort called Project Salus to combine dozens of sources of data. From that data, the center could project potential shortfalls of key supplies such as food, water, medical equipment, protective gear and other essentials.
That effort, first reported by CQ Roll Call in April, has led to more than 70 different data sources being brought together — including government agencies such as the Census Bureau and Medicare, as well as private entities such as retailers, hospitals and others — to produce models that can predict shortfalls of food or medical needs across the country as the pandemic spreads, Shanahan said.
The Pentagon has provided early versions of the models to the U.S. Northern Command, which uses them as part of its common operating picture, Shanahan said.
The command is responsible for defending the continental United States and territories and provides military aid to non-military agencies such as the Federal Emergency Management Agency.
The command has deployed Army and Navy medical personnel to New York, sent Navy hospital ships to New York City and Los Angeles, and helped set up field hospitals in areas where local health care facilities were overwhelmed with patients.
Northern Command has said it has been working with several top U.S. technology companies — including Apple, Microsoft, mapping software maker Esri and Monkton, a company that helps developers build secure apps for classified purposes — to help FEMA and other agencies during the pandemic.
Although the Pentagon has faced skepticism from some tech companies in its pursuit of artificial intelligence technologies, and outright refusal by Google to continue collaborating on a Pentagon project to identify and label objects in drone videos, the pandemic appears to have changed the calculation, Shanahan said.
As soon as the Pentagon launched Project Salus to build predictive models of shortages during the pandemic, there has been “an outpouring of support from private companies,” as well as major universities, Shanahan said. The top tech companies in the country and their teams of artificial intelligence and machine learning specialists have shown a “strong desire” to work with the Defense Department, he said.
JAIC declined to name tech companies involved in the effort because the work is still in prototype form, and the companies have not agreed to be named publicly. Shanahan said he hoped that the current collaboration between the Pentagon, tech companies and academia would recreate some of the past successes of such joint ventures.
Previous ventures involving the Defense Department, Bell Laboratories and the Massachusetts Institute of Technology had produced many of the technologies that undergird the internet and gave birth to Silicon Valley, he said. “We have lost that in the last 20 years,” Shanahan said.
The pandemic also could reshape defense budgets in the years to come as Congress has approved about $3 trillion toward economic relief measures.
“I don’t see any possible future other than lower budgets for the department over the next few years because of what has been spent,” Shanahan said. “You cannot keep going down the path of buying everything you had before and still try to get into the digital future.”
More of the Pentagon’s budget should go into accelerating artificial intelligence-based systems and technologies and less into industrial-age hardware, Shanahan said.